accelerometer driver

[Wolfgang Spraul]

Harald,
I have heard a lot of mysteries around the accelerometers.
Are they very sensitive or not? How sensitive? Are they calibrated  
somewhere?
In the factory we saw big variations of values they produced, when  
lying flat on a table. Nobody could really understand why.
Do we need 2 accelerometers, if they are already 3D? We are planning  
to take one out in future products... What do you think?
Until today there are bugs in the driver talking to both at the same  
time I think, in other words even if there is some advantage to having  
two, they are not working right yet...
Just fyi,
Wolfgang

On Aug 29, 2008, at 11:49 AM, Harald Welte wrote:

> On Thu, Aug 28, 2008 at 02:06:51PM +0900, Andy Green wrote:
>
>> I don't know of one with mainline support off the top of my head.
>
> I am not aware of any mainline acceelerometer drivers.
>
>> The LIS302DL threshold stuff will be useful when it is  
>> implemented.  A
>> lot of alternative motion sensors are analogue type, which we can use
>> fine with CPU ADC, but then we don't get the power saving from  
>> threshold
>> stuff.
>
> I originally voted strongly against analog accelerometers, since  
> then you have
> to route very sensitiva analog signals half across the PCB.  The  
> accelerometers
> have to be at a maximum physical distance from each other, i.e.  
> there will be
> long traces with fragile analog signals.
>
> Also, each analog accelerometer has three analog signals (one for  
> each axis),
> which is six signals, i.e. more than our s3c24xx ADC has input  
> channels.
>
> Also, having irq-based tap / double-tap / free-fall detection in  
> hardware
> seemed really neat to me.
>
> -- 
> - Harald Welte <laforge at openmoko.org>          	        http://openmoko.org/
> = 
> = 
> = 
> = 
> = 
> = 
> ======================================================================
> Software for the world's first truly open Free Software mobile phone